Stanford CS547 HCI Seminar | Autumn 2026 | Sensing Uncertainty in Sustainable HCI
Summary
This seminar explores the critical intersection of sustainability and Human-Computer Interaction (HCI), focusing on 'sensing uncertainty.' The core argument addresses the 'low-carbon innovation paradox,' where technologies designed to solve environmental issues (like IoT) may inadvertently increase resource consumption. The talk proposes that instead of focusing on near-term vs. long-term benefits, designers must embrace uncertainty in the data itself. Two major case studies are presented: 1) Upcycling, which requires embedding computing in secondhand, materially complex objects, and 2) Air quality data visualization, which demonstrates how user interfaces must communicate the range of possible values rather than a single point estimate.
Key takeaways
-
The Low-Carbon Innovation Paradox
2:20
Technologies like IoT, while promising for resource management, increase the ratio of devices per person (potentially 50-100 devices per person in built environments), contributing to carbon emissions through manufacturing, networking, and computation.
-
Upcycling and Material Risk
10:00
Integrating computing into secondhand materials requires novel design approaches that account for material traces, structural inconsistencies (e.g., embedded metal, toxic coatings), and the need for expert, hands-on knowledge.
-
Ubench System Design
17:00
The Ubench system supports upcycling by combining hands-on engagement with sensing tools. It uses a fine-tuned segmentation model to detect wood contours and defects (cracks, knots), and instrumented sensing pins (inductive/capacitive) to detect hidden risks like metal or moisture, projecting annotations directly onto the workpiece.
-
Communicating Data Uncertainty
27:30
When data is uncertain (e.g., air quality readings), visualizations must communicate the possibility that an event did not occur as reported. The talk tested static, animated, and interactive visualizations, finding that a trade-off exists between supporting accurate interpretation (static) and understanding data reliability (interactive).
Technical details
-
Ubench System Architecture
1020s
A system designed for upcycled IoT, Ubench combines hands-on engagement with sensing tools. It uses an overhead projected camera unit for assessment and an in-situ design model. Key components include: 1) A fine-tuned segmentation model for visible defects (cracks, knots). 2) Instrumented sensing pins (e.g., inductive sensor for hidden metal, capacitive sensor for moisture) tracked in 3D using fidial markers. 3) An interactive dialogue that provides contextual suggestions when design intent intersects with material risk.
-
Air Quality Data Visualization
1650s
The research used the EPA's guidance to map PM2.5 concentrations to the Air Quality Index (AQI). Uncertainty was visualized across three dimensions: 1) Static (using color/glyphs), 2) Animated (using animated hypothetical outcome plots), and 3) Interactive (with a slider). Spatial context was handled using the Kriging algorithm to interpolate data among sparsely placed sensors.
-
Data Processing and Modeling
1750s
The analysis utilized the Root Mean Square Error (RMSE) to measure and encode uncertainty. The study employed a within-subjects design to test user comprehension of data reliability and accuracy across different visualization types and contexts (spatial, temporal, sensor-only).
Mentioned resources
Channel & topics
Watch on YouTube · Back to latest
This independent, AI-assisted summary is provided for commentary and informational purposes. It may contain errors or omit important context. Please watch the original video for the creator's complete presentation. Video, thumbnail, and related copyrights belong to their respective owners.